atsc 3.0 broadcast 5g unicast heterogeneous network ......2020/07/13 · the 5gc architecture is in...

This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

IEEE TRANSACTIONS ON BROADCASTING 1

ATSC 3.0 Broadcast 5G Unicast HeterogeneousNetwork Converged Services Starting Release 16

Michael Simon , Ebenezer Kofi, Louis Libin, Member, IEEE, and Mark Aitken , Member, IEEE

Abstract—In this paper, the ATSC 3.0 broadcast Radio AccessTechnology (RAT) is aligned with 3GPP 5G NR RAT in the con-text of 5G convergence starting in Release 16. The 5G systemarchitecture release 16 includes a new 5G physical layer knownas 5G NR “New Radio” and a “Cloud Native” 5G Core (5GC)using cloud computing. The 5GC is agnostic to the type ofradio access technology used and is enabler of many typesof convergence. A novel shared multi-tenant broadcast corenetwork architecture designed to interwork with 5GC is dis-cussed. With the 3GPP 5G NR unicast and Non-3GPP ATSC3.0 broadcast synergistically aligned using methods of Release16. This includes using 3GPP Access Traffic Steering, Switching,Splitting (ATSSS) and a multi-radio dual simultaneous connectedUser Equipment (UE). This aligns ATSC 3.0, the first forwardlooking (non-backward-compatible) native IP OFDM broadcaststandard, with 3GPP LTE/5G unicast as a converged 5G vertical.The proposed method and architecture are orthogonal to LTEbroadcast Release 16 and is synergistic to 5G NR mixed-modemulticast unicast in future.

Index Terms—5G core, 5G NR, ATSC 3.0, broadcast core,ATSSS, heterogeneous network convergence.

I. INTRODUCTION

THE FUTURE of wireless systems operational architec-tures such as in 3GPP 5G is profoundly changing by

the adoption of new radio technologies driven by virtualizedcore networks using cloud computing. The 5G architecturein Release 16 satisfies requests for each service use case byorchestrating the chaining together of virtual network func-tions to create independent services as end-to-end logical slicesof the physical 5G network and this paradigm shift is termed5G network slicing.

The current USA broadcast regulatory environment per-mits broadcasters to voluntarily innovate using the ATSC3.0 physical layer standard with their spectrum for newservices including mobile. It permits shared use of licensedbroadcast spectrum and transmission infrastructure. Therefore,a new multi-radio dual connected user equipment (UE) using3GPP 5G NR and Non-3GPP ATSC 3.0 broadcast synergis-tically aligned is proposed in this paper. 3GPP has defined

Manuscript received December 19, 2019; revised February 4, 2020and March 13, 2020; accepted March 26, 2020. (Corresponding author:Michael Simon.)

The authors are with the Advanced Technology, Sinclair Broadcast Group,Hunt Valley, MD 21030 USA (e-mail: [email protected]; [email protected];[email protected]; [email protected]).

Color versions of one or more of the figures in this article are availableonline at http://ieeexplore.ieee.org.

Digital Object Identifier 10.1109/TBC.2020.2985575

a 5G NR (New Radio) [1] a cloud-native Service BasedArchitecture (SBA) 5G Core (5GC) [2], [3], [4].

The 5G NR operates in multiple 3GPP spectrum bands anduses 3GPP and Non-3GPP RAT with dual connected UE forthe service requirements of 5G in Release 16.

Multi-band, multi-RAT dual connected UE using both3GPP and Non-3GPP RAT is specified in 3GPP TS 22.261[5, Sec. 6.3]. A new Non-3GPP ATSC 3.0 broadcast RAT isproposed that is synergistically aligned to 5G NR by usingthe existing methodology specified in Release 16. This canbring valuable low-band spectrum into play that can offer newaligned broadcast 5G converged use cases.

The ATSC 3.0 broadcast standard [6], [7], [8] is the firstand only forward looking (non-backward-compatible) OFDMbroadcast physical layer standard using a native IP trans-port, designed for extensibility and future alignment withthe 4G LTE physical layer. The ATSC 3.0 physical layeris now also aligned to 5G NR physical layer frame andL1 signaling to enhance 5G convergence while using 3GPPATSSS [12], [13], [24].

This paper proposes innovative change by aligning Non-3GPP ATSC 3.0 broadcast RAT with 3GPP 5G NR unicastRAT. The unicast RAT and broadcast RAT are first eachoptimized for their diverse radio physics environments andreceived using a dual simultaneously connected UE.

The current 3GPP spectrum bands for 5G are divided intolow-band sub-1 GHz, mid-band 1- 6 GHz and high-band above6 GHz. The low-band spectrum physics, enables operator bothlarger cell coverage and deeper in-building penetration, etc.The mid-band offers reasonable coverage and good capac-ity with bandwidths up to 100MHz. The high-band withbandwidths up to 400MHz and with massive MIMO and beam-forming can increase capacity and efficiency. However, thedisadvantage to high-band is reduced cell size and the signal’ssusceptibility of being blocked by objects in the environment.

The clear premise is that spectrum is not fungible givenradio physics. Therefore, the 5G NR system architecture inRelease 16 is designed to support the synergy of a multi-band,multi-RAT heterogeneous network (HetNet) architecture anda dual simultaneously connected UE.

The remainder of this paper is structured as follows:Section II describes several 5G Release 16 deployment systemarchitecture options analogous to proposed broadcast 5G con-vergence system architecture; Section III compares some rele-vant 5G NR and ATSC 3.0 RAT aspects; Section IV describesthe proposed ATSC 3.0 RAT alignment with 5G; Section Vintroduces the 5G Core in the context of supporting 3GPP and

0018-9316 c© 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See https://www.ieee.org/publications/rights/index.html for more information.

Authorized licensed use limited to: Michael Simon. Downloaded on April 25,2020 at 23:41:33 UTC from IEEE Xplore. Restrictions apply.

https://orcid.org/0000-0003-0881-8811

https://orcid.org/0000-0003-1828-7057


2 IEEE TRANSACTIONS ON BROADCASTING

Fig. 1. (A) 5G NSA LTE as Anchor, (B) 5G NSA 5G as Anchor.

Non-3GPP access; Section VI introduces a proposed BroadcastCore interworking with 5G Core; Section VII describes exam-ple use case USA of shared Broadcast Core and platform for5G Convergence; Section VIII provides conclusions and futureplanned research.

II. SEVERAL 5G RELEASE 16 DEPLOYMENT OPTIONS

Figure 1 depicts a high-level view of several 5G Non-Stand Alone (NSA) architecture deployment options. Theseare briefly introduced for context and to improve initial under-standing of the methodology in Release 16. These options havesimilarities to the proposed aligned broadcast 5G convergencesystem architecture shown in Figure 2.

Figure 1 (a) is NSA option 3x, a popular choice for earlydeployment of 5G NR for achieving higher data capacity. This5G deployment option is termed Non-Stand Alone because the5GC is not deployed, but 5G depends on LTE which is termedan anchor. In option 3x, the MNO uses existing 4G LTE basestation (eNB) and Evolved Packet Core (EPC) as anchors for5G NR and multi-RAT dual connected UE in Release 16.

Figure 1 (a) the 4G LTE eNB is termed Master Node (MN)or cell and uses existing low-band spectrum. The LTE EPCserves as an anchor to support the 5G gNB SecondaryNode (SN) or cell in mid-high band spectrum for higherdata capacity.

An essential detail of Figure 1 (a) is that both the eNB andgNB has an independent OFDM MAC layer scheduler basedon a specific standard 4G LTE, 5G NR, etc. These sched-ulers then allocate OFDM resources in the normative mannerexpected by UE. Some details Figure 1 (a) are that the eNBMN and LTE RAT 1 both have user plane (U) and controlplane (C) data from EPC shown. The EPC and eNB then serveas an anchor for the control plane (C) of 5G RAT2 as shownbetween LTE RAT 1 and 5G RAT 2 in the UE. The gNB isa SN and only has (U) shown provided either from the EPC toRAT 2 or via an interface shown between the eNB and gNBto 5G RAT 2.

An advantage of using this HetNet with 5G NR in high-band spectrum and given the knowledge that the Up Link (UL)

Fig. 2. Broadcast 5G convergence architecture proposed.

from 5G UE is the limiting factor for 5G NR coverage. InNSA option 3x, the EPC can seamlessly schedule UE to usethe LTE eNB for the 5G UL on the low-band spectrum. Thiseffectively extends the 5G NR coverage area using the physicsattributes inherent in the low band spectrum.

Another HetNet attribute used to mitigate early 5G NRdeployment in the mid-high band is by improving the con-tinuity of service perceived by the consumer. When the dualconnected UE roams outside the small cell 5G NR coverage,LTE 4G can continue to provide seamless service using largerLTE 4G low-band coverage area in a way that is agnostic tothe consumer.

In Figure 1(b) NSA option 7x is shown. 5GC is usedinstead of EPC supporting 5G NR for high data capacity.Moreover, the 5GC cloud computing architecture is requiredto orchestrate all new 5G use cases and verticals, with highdata capacity and needed intelligence including on the edgeof the network for low latency uses cases and 5G networkslicing. In Figure 1 (b) the roles of MN and SN are reversedfor the eNB and the gNB and the 5GC and gNB are usedas an anchor for 4G LTE RAT 2. Again, the diverse HetNetattributes are used to improve service quality in a way agnosticto the consumer.

Figure 2 is high-level view of the proposed broadcast 5Gconvergence architecture using methodologies of Release 16 inthis paper. In this proposed HetNet architecture, the dualconnected UE has 3GPP 5G unicast RAT 1 and Non-3GPPbroadcast RAT 2. The ATSC 3.0 base station is SN with (U)only from broadcast core (BC) network. The 5G RAT 1 hasboth (U) and (C) from the gNB as MN, and 5GC. The 5GC andthe BC user planes (U) and control planes (C) interwork usingthe interfaces shown. The ATSC 3.0 broadcast RAT 2 uses thegNB MN and 5GC as anchor for (C) and Uplink (UL) in thisconverged architecture which is discussed in more detail inSection VII.

III. 5G NR AND ATSC 3.0 RAT ASPECTS

The diverse nature of unicast and broadcast physics whenacknowledged and accepted can be synergistically leveraged



SIMON et al.: ATSC 3.0 BROADCAST 5G UNICAST HETEROGENEOUS NETWORK CONVERGED SERVICES STARTING RELEASE 16 3

as a net positive for efficiency, performance and new use casesin 5G heterogeneous networks using Release 16 methodology.

The central defining distinction between the 5G unicastdesign philosophy compared to ATSC 3.0 broadcast designphilosophy is that 5G unicast has both a downlink and returnchannel uplink and heavily depends on using adaptive mod-ulation coding scheme (MCS) and error recovery protocols.The 5G unicast scheduler receives reports on channel condi-tions at UE. The 5G unicast scheduler then selects modulationcoding scheme (MCS) and OFDM sub-carriers or resourceblocks with best channel conditions at UE. Moreover, MCS isselected for efficiency, knowing availability of adaptive MCSfeedback UE and L2 layer MAC, RLC, PDCP protocols thatoffer re-transmission of uncorrected errors at UE.

Also, a relevant subject is the legacy 3GPP method usedto combine broadcast OFDM resources into some sub-framesand unicast OFDM resources in other sub-frames of a shared10ms unicast frame to introduce broadcast services. Thesame general approach is now and termed “Further evolvedMultimedia Broadcast Multicast Service” (FeMBMS) and isused LTE broadcast in Release 16 with focus mostly on fixedbroadcast television services. There is also a new 5G NRmixed-mode multicast unicast in 10ms 5G NR frames for thefuture [25]. These are both orthogonal to the method disclosedin this paper.

The LTE legacy method used to introduce broadcast serviceshas deep rooted constraints. The first, is by not respecting thediverse physics of unicast and broadcast. Second, is the timemultiplexing of both unicast and broadcast OFDM resources ina shared 10ms unicast frame optimized for only unicast. Thisresults in a broadcast RAT lacking essential time diversity andefficiency. The unicast design philosophy views time diversityas adding latency and is not valued. However, using broadcastdesign philosophy where latency is not a primary constraint,time diversity greatly boosts performance and efficiency.

The ATSC 3.0 broadcast design goals are very different,understanding that the emitted signal purpose is to be receivedby many UE simultaneously without knowledge of channelconditions at any UE. The design of the ATSC 3.0 RAT usesan optimized Low-Density Parity Check (LDPC) and Non-uniform QAM (NU-QAM) constellations resulting in spectralefficiency that closely approaches the Shannon limit [9].

Additionally, ATSC 3.0 physical layer design philosophyoffers multiple forms of diversity to improve performance andefficiency by helping to mitigate impulse noise and mobilefading channels, etc. The ATSC 3.0 diversity domains include:

A. Frequency domainB. Time domainC. Power domain (LDM)D. Transmitter spatial domainE. Channel Bonding (frequency domain)F. UE antennas SISO spatial diversity (MRC)(A) First, frequency diversity is achieved directly at the

physical OFDM symbol by the sub-carriers carrying contentdata being interleaved across the entire channel bandwidthbefore emission with the inverse process applied at the UE.

(B) Next, time domain diversity is applied to the outputcells of the LDPC non-uniform constellation block. These cells

represent user plane IP data flows and are termed a physi-cal layer pipe (PLP). These PLP cells are time interleaved toa depth of either: 50ms, 100ms, 200ms and then use dispersedPLP mapping into non-contiguous resource blocks across thesymbols of a sub-frame. The broadcast frame size is selectablefrom 50ms-5000ms and with the dispersed PLP mapping pro-vides additional time diversity by spreading the already timeinterleaved PLP cells out further in time in the sub-frameto improve impulse noise and the mobile fading channelperformance. For example, note that in [10], [22], the mobilebroadcasts show irrefutable improvements using time diversityfor ATSC 3.0 mobile fading performance and efficiency.

(C) Transmitter spatial diversity is termed a single frequencynetwork has multiple synchronized coherent broadcast trans-mitter sites emitting identical signals that are received fromdifferent spatial directions at the UE. This spatial diversityhelps to mitigate obstructions (shadowing) in the propagationpath and fading at UE receive antenna. These multiple coher-ent signals are then combined for gain at UE and are a majorfactor in the increased broadcast performance and efficiency.

(D) Non-Orthogonal Multiple Access (NOMA) in the powerdomain is termed Layer Division Multiplexing (LDM) inATSC 3.0 [11]. This offers another domain with many usecases including combined unicast and broadcast services [21].

(E) ATSC 3.0 channel bonding uses separate carriers toeither increase capacity similar to 4G/5G carrier-aggregation.Another option is bonding two synchronized carriers thatare time multiplexed on a cell-by-cell (sub-carrier) basisfor increased robustness via frequency diversity across non-contiguous carriers bonded in the broadcast band.

(F) For ATSC 3.0 SISO UE antenna diversity usingMaximum Ratio Combining MRC is beneficial and efficient.

IV. ATSC 3.0 RAT CONVERGENCE 5G ALIGNMENT

The ATSC 3.0 specialist group TG3/S32 designed the RATto be flexible and extensible, an essential ATSC 3.0 systemrequirement. The OFDM numerology is defined by:

�F (Hz) = Fs (Hz)

IFFT (size)(1)

where �F is sub-carrier spacing (Hz), and Fs is OFDM sam-pling frequency (Hz). The Fs (Hz) was carefully selected forATSC 3.0 RAT so as not to preclude future alignment tothe 3GPP RAT. This was technically assured by calculatedselection of the ATSC 3.0 sampling frequency Fs:

Fs := [N] × 384,000 Hz (2)

The factor 384,000 relates to WCDMA chip rate and wasimportant in 4G LTE for compatibility reasons. The same Fs =[N] x 384,000 Hz (2) is used in ATSC 3.0, 4G LTE and5G NR.

Table I is a consolidated view of the OFDM numerologyin standards: 4G LTE, 5G NR and ATSC 3.0 respectfully andshows the value [N] selected and some OFDM parameters forcomparison.

In ATSC 3.0 standard the value [N] is signaled to UE using7 bits in the A/321 L1 signaling in each frame. This ensuresextensible OFDM numerology for future frames designed for




Fig. 3. ATSC 3.0 Frame Lengths (A) 80ms, (B) 160ms, (C) 10ms Time Aligned 5G NR 10ms Frames and Synchronized SFN counts.

TABLE ICONSOLIDATED VIEW OFDM NUMEROLOGY LTE 4G, 5G NR, ATSC 3.0

new specific use cases such as mobile or IoT, etc. The ATSC3.0 RAT is designed to time multiplex different frames withdifferent OFDM numerology and using L1 signaling so thatUE gracefully ignores frames it is not capable of receiving.

In this paper, this related numerology is used as a basis totime align ATSC 3.0 frames to 5G NR unicast frames emittedat air interfaces of respective downlink antennas using the TAIreference clock. Also, the L1 signaling in the ATSC 3.0 framehas a System Frame Number (SFN) count phase locked to theSFN count in 5G NR unicast L1 signaling and these are bothavailable at dual connected UE.

In 5G NR, the SFN count is based on 10-bit mod1024 counter incremented with every 10ms frame. TheSFN count wraps to zero every 1024 (frames) × 10ms =10,240ms.

In ATSC 3.0, the frame length is selectable 50ms to 5000ms.Then, to enable 5G NR alignment, the ATSC 3.0 frame lengthis selected ensuring an integer number ATSC 3.0 frames occur

in the 10,240ms SFN count period using:

Frame Length = 1024

2Nx10ms[N = 2, 3, 4, 5, 6, 7, 8, 9, 10]

(3)

This results in the possible aligned ATSC 3.0 frame lengths:10ms, 20ms, 40ms, 80ms, 160ms, 320ms, 640ms, 1280ms,2560ms, with 1024, 512, 256, 128, 64, 32, 16, 8, 4, or 2 ATSCframes respectively in the 10,240ms SFN count period.

Figure 3 shows an example of time alignment of 10ms,80ms, and 160ms ATSC 3.0 frames to 5G NR 10ms framesand with all L1 signaling SFN counts phased locked.

Then, to first establish and or then maintain frame alignmentand synchronized SFN counts requires use of a TAI GPS ref-erence clock shown. The synchronization algorithm uses GPSepoch 1980-01-06 at midnight UTC. The GPS epoch definesthe instant that the first 5G NR and ATSC 3.0 frame wereemitted antenna air interfaces with SFN counts equal to zero.To calculate SFN count at any instance in the future use:

SFN count = GPS seconds × 100 mod1024 (4)

Then GPS seconds since epoch available from a GPSreceiver x 100 converts seconds to 10ms increments, then mod1024 establishes phase of SFN count (0-1023) at any instanceand at any node in respective networks with GPS available.

The frame length (3) selection can be changed seamlessly on10,240ms boundaries when using either 5G NR Time DivisionDuplex TDD or Frequency Division Duplex FDD with TDDused more in 3GPP mid-high band spectrum, and once thealignment is established, it will not drift using TAI reference.

Table II shows one example of future extended numerol-ogy using (1) and values [N] selected to optimize for mobilebroadcast 5G convergence to be discussed in Section VII.

Table II shows some relevant frequency ranges of 200 –700 MHz and BW 5-20MHz in context of example to be




Fig. 4. 5G Core cloud-native architecture abstracted and context of supporting both 3GPP and Non-3GPP access networks.

TABLE IIATSC 3.0 OFDM NUMEROLOGY EXTENDED FOR MOBILE

discussed in Section VII. In last two columns is �F (Hz) andmobile doppler performance km/Hr. Note: new �F (Hz) val-ues and the new 4096 IFFT size as compared to Table I thatare optimized for fixed television service, but now extendedfor mobile. It should be appreciated with 7 bits in L1 signalingthe value [N] (3) this is only one example of the extensibilityavailable.

V. 5G CORE INTRO CONTEXT 3GPP AND

NON-3GPP ACCESS

Figure 4 is the new 5GC with a cloud-native Service BasedArchitecture (SBA) [14] and with several relevant functionsdiscussed. The paradigm shift brought by the 5GC comparedto legacy 3GPP standards is the method used for separation of

control and user plane in the core network shown in Figure 4.The control plane entities are instantiated as software func-tions shown running in the cloud environment using methodsof large IT players by using a SBA. The dashed lines areinterfaces connecting all control plane functions.

The 5GC architecture is in great contrast to the past stan-dards such as GSM, WCDMA and LTE which each haddifferent core networks, radio access networks and used dif-ferent protocols, etc. This monolithic nature ends with 5GCdesigned to be abstracted (non-aware) of access network,3GPP or Non-3GPP.

The control plane functions are no longer static in theirfunction. Instead, each of the functions shown in Figure 4must first register with Network Repository Function (NRF)entity shown using interface and APIs defined. Then, onceeach control function’s services are registered in NRF, theybecome available to the other registered functions by discov-ery via the NRF. This flexibility allows dynamically chainingtogether functions to build end-to-end services and this istermed 5G network slicing and uses the methodology of largecloud players such as Facebook, Amazon, Netflix, Google, etc.

This was a deliberate decision by 3GPP to increase flexibil-ity of 5G system, to be innovative and to remain competitiveusing methodology of large cloud players. This allows the5G NR data capacity to be translated into increased revenues,competitive services and enable new 5G industry verticals.

The 5GC user data plane (U) runs from the Internet ordata network via N6 interface to the User Plane Function (UPF)shown and then via N3 interface to 3GGP radio accessnetwork and using N3 to Non-3GPP radio access networkusing Non-3GGP Interworking Function (N3IWF) shown.




Fig. 5. The 5G Core Interworking with the Broadcast Core in a Next Gen Wireless Platform (NGWP).

The Access Mobility Function (AMF) interacts with the3GPP radio access network and N3IWF using N2 interfaceand the UE via N1 interface respectively. The SessionManagement Function (SMF) via N4 to UPF manages all ses-sions and allocation of IP addresses, etc. The AuthenticationServer Function (AUSF) authenticates the UE. The UnifiedData Management function (UDM) stores subscription dataand generates authentication data for AUSF. The UnifiedData Repository (UDR) is a generic database enabling state-less operation of all functions in cloud for flexibility andresiliency to support interworking with other core networks andexternal entities, the 3GPP has defined the Network ExposureFunction (NEF). This gives access or controlled exposure of5GC network core functions to external entities and is alsoused in 5G vertical use cases, and broadcast 5G convergenceSection VII.

VI. 5G CORE BROADCAST CORE INTERWORKING

Figure 5 introduces high-level view 5GC interworkingwith a shared Broadcast Core (BC) to prepare for dis-cussing the proposed broadcast 5G convergence architecturein Section VII. The BC in Section VII is an integrated part ofa Next Gen Wireless Platform (NGWP), which is a multi-tenant system platform shared by licensed broadcasters toconverge broadcast services with 5G MNO using Release16 methodology.

The convergence is coordinated both at IP layer interwork-ing (5GC and BC) and with aligned 5G NR and ATSC 3.0 RATschedulers and with ATSSS rules in core networks and at thedual connected UE shown. The 5GC is used as anchor for ULATSC 3.0 broadcast which has only DL and (U) as first shownin Figure 2.

The BC shown in Figure 5 is a cloud-native architec-ture and uses some of the control plane functions discussed

in Figure 4 for the 5GC. The BC is shown to the rightof interworking interfaces for the control and user planesshown. Several blocks, such as BC orchestration, spectrumpools, Broadcast Virtual Network Operator (BVNO) and thecloud interfaces shown are a part of the NGWP discussed inSection VII.

The interworking and ATSSS functions (AT3SF) are shownas greyed blocks and their functionality will be briefly dis-cussed. The 5GC control plane interworking entity is theNEF as was discussed and defined in [2]. The BC controlplane interworking entity is the Broadcast Network ExposureFunction (BNEF). The 5GC user plane interworking entity isthe N3IWF, and the BC user plane interworking entity is theUPF as shown.

For normal (non-interworking) operation, the 5G NR userplane IP data flow is from the Internet or data networkvia N6 to UPF then via N3 to the gNB in 3GPP RAN. The nor-mal (non-interworking) operation for broadcast is from BVNOto broadcast UPF and then ATSC 3.0 scheduler and then ATSC3.0 exciter in Non-3GPP broadcast RAN as shown in Figure 5.

The interworking between 5GC and BC provides the bi-directional IP flows to support the converged use of unicastand broadcast at IP layer in core networks and at the dualconnected UE using ATSSS Release 16.

The 3GPP ATSSS Release 16 describes the access trafficsteering, switching or splitting by using these definitions [16].

Steering is the procedure that selects an access network fora new data flow and transfers all traffic of this data flow overthe selected 3GPP or non-3GPP access network.

Switching is the procedure that moves all traffic of an ongo-ing data flow from one access network to another accessnetwork in a way that maintains continuity of the data flowusing 3GPP and non-3GPP access networks.

Splitting is the procedure that splits data flow acrossaccess networks. With some traffic of data flow transferred




via 3GPP and other via a Non-3GPP access networksimultaneously.

In this paper, the ATSSS functionality is used and discussedin the context of the proposed broadcast 5G convergencearchitecture. Currently, in 3GPP ATSSS, the non-3GPP accessnetwork can be Wi-Fi 6 [23], fixed access network [17],satellite [18] or as here proposed, ATSC 3.0 broadcast.

The ATSSS functionality is determined by the core networkpolicy established on PCF, PC-AT3SF, and by the ATSSSrules distributed as shown to the user plane and control planeAT3SF functions in 5GC and BC. The ATSSS rules are thendistributed to dual connected UE. The ATSSS switching func-tionality can be supported using MPTCP protocol [26] aboveIP layer or ATSSS-LL at lower layer on the UPF shown inBC for broadcast multicast IP flows.

When the dual connected UE shown in Figure 5 attachesto the 5GC, the UE announces to the 5GC its capabili-ties including a broadcast Non-3GPP RAT (ATSC 3.0) underATSSS. Then with ATSSS policy and rules distributed in the5GC, BC, and UE. The network conditions such as conges-tion 5GC or the signal conditions at UE are used by ATSSS toimprove network efficiency and QoS for users in an agnosticway. The 5GC and BC typically interwork under a ServiceLevel Agreement (SLA), between broadcaster (BVNO) andMNO, and will be discussed in Section VII.

The interworking provides for managed bi-directional IPflows for the explicit converged use of unicast and/or broad-cast services at the dual connected UE with ATSSS Release16. In Figure 5 the IP data flow in 5GC can be directedunder ATSSS from UPF in 5GC to N3IWF to UPF in BCto the ATSC 3.0 scheduler to the Non-3GPP broadcast RANand finally the broadcast signal is received at dual connectedUE. Conversely, in the BC the IP data flows can be directedfrom UPF in BC to N3IWF and then to UPF in 5GC thenvia N3 to 3GPP RAN and gNB, and finally unicast signals arereceived at dual connected UE using ATSSS rules and deci-sions made in core networks and dual connected UE based onreal-time conditions.

In 3GPP ATSSS Release 16, the user plane switching func-tionality is supported in 5GC and BC using the UPF and atthe dual connected UE. The MPTCP protocol can be used forTCP/IP flows and the ATSSS-LL (Lower Layer) can be usedfor any IP flow including UDP, TCP, etc.

Both ATSSS switching methods MPTCP and ATSSS-LLcan be used simultaneously and selected on an IP flow basis onUPF and at the dual connected UE. Then, when dual connectedUE attaches to the 5GC, it announces its capabilities and selec-tions under ATSSS rules established. For ATSC 3.0 broadcastthe ATSSS-LL method is used only in BC on UPF and at UEwith middle-layer IP stack support for UDP/IP multicast IPflows [2].

VII. PROPOSED SHARED BROADCAST CORE AND NGWP

Effective March 5, 2018, the FCC adopted portions of theATSC 3.0 broadcast standard (A/321, A/322) and permitedits use on a voluntary, market-driven basis in the UnitedStates [16]. These FCC rules permit broadcasters voluntar-ily deploying ATSC 3.0 to enter private business agreements

to share their 6 MHz broadcast channel and use a com-mon transmission infrastructure to innovate using the ATSC3.0 standard. However, the constraint is the ATSC 3.0 standardlacks both a method and system architecture for the efficientsharing of broadcast spectrum and transmission infrastructure.

The NGWP architecture is shown in Figure 6 that enablesbroadcasters in the U.S. to share spectrum efficiently usinga virtualized Radio Access Network vRAN infrastructure toinnovate using ATSC 3.0. Also, each broadcasters can havethe option to explore new broadcast 5G converged services andmobile business models using the extended OFDM numerol-ogy discussed in Table II. The NGWP is a SDN/NFV cloudnative multitenant platform environment optimized to enablethe broadcast spectrum to be shared for ATSC 3.0 services.The NGWP includes integrated BC that provides the for eff-cient automated spectrum sharing and the use of a commonbroadcast RAN infrastructure.

The NGWP virtualizes the broadcast spectrum as agreedby the broadcasters through private sharing agreements andthis is reflected in Broadcast Market Exchange (BMX) pol-icy running in NGWP. The shared spectum resources usage isaccounted for by the establishment and management of a spec-trum resource pool for each ATSC 3.0 6 MHz channel beingshared in NGWP.

A broadcaster using the services of NGWP in this paperis termed a Broadcast Virtual Network Operator (BVNO).A BVNO is shown at top of Figure 6 sending IP flowsfrom an external automation playout environment into NGWPusing the northbound interface. The BVNO is authorized andauthenticated by the BMX orchestration entity which has thepolicy running reflecting the specfics of the spectrum sharingagreements of all BVNO.

The NGWP has edge datacenters shown each witha Spectrum Resource Manage (SRM) entity that controls (N)ATSC 3.0 schedulers to build (N) 6 MHz ATSC 3.0 framesinstructed by BMX orchestration entity under multi-tenantsharing. The ATSC 3.0 scheduler creates digital basebandframes that are sent via NGWP southbound interface to theATSC 3.0 exciter in the shared RAN environment, which con-verts digital baseband to an analog RF signal for broadcastinto spectrum. This supports the several coherent transmittersof a single frequency network for benefit of transmitter spatialdiversity discussed Section III. The signal is then shown beingreceived by a commercially available ATSC 3.0 receiver forfixed services selected by the BVNO.

The SRM entity creates a broadcast frame record for eachframe and this is used to validate each tenants usage of spec-trum resources in real-time based on BMX policy for thelicensed local geographic area served from edge datacenter.This could also form a national platform using the CognitiveSpectrum Management (CSM) entity shown in Figure 6 havinga consolidated global view of all regional edge datacen-ters. The CSM abstracts the details of spectrum pools fromand provides service to the BMX orchestration entity shown.The BMX orchestration entity has the business view, policy,charging, etc. to monetize spectrum resources in all pools.

The term Broadcast Market Exchange BMX relates to theoption of treating the BVNO spectrum resources as commodi-ties in a market exchange. Then each BVNO has the option to




Fig. 6. Converged 3GPP 5G NR and Non-3GPP ATSC 3.0 Services in 5GC / BC and at Dual Connected UE ATSSS release 16.

dynamically sell or buy additional spectrum resources usinga BMX with other participating tenants. This automation setsthe stage to put broadcast spectrum into play for innovativemobile broadcast 5G convergence business models with a 5GMNO as another tenant interworking under SLA in the future.

The 5GC and BC interworking entities previsiouly discussedin Figure 5 are now shown in Figure 6 with IP data flowsATSSS switched in the UPF entities connecting either the3GPP or Non-3GPP RAN.

The attention is now turned to the timeline shown at the bot-tom of Figure 6 of the time aligned frames with synchronizedsystem frame number counts locked to TAI, shown from per-spective of the dual connected UE. The timeline at points (X),(Y), (Z) shows the result of the coordinated unicast and broad-cast RAT schedulers under ATSSS Release 16 which can besimultaneously active on UE.

Then, using ATSSS policy, rules and conditions in thecore networks and with reports from the UE via PerformanceMeasurement Function (PMF) (not shown) determines theATSSS functionality to improve network efficiency and qualityof service to consumer in an agnostic way. Using AT3SF, theUPF is instructed, and by using interworking interface, cansteer or switch IP traffic flows to the selected RAN. Also,ATSSS splitting enables same IP flow to be split to both3GPP and Non-3GPP RANs and received by dual connectedUE shown.

As shown at SFN count instant (X) is an example of IP flowfor a service using only the unicast RAT. Then at SFN countinstant 0016 (Y) the IP data flow for the service is switchedwith only the broadcast RAT active. Then, continuing in timeto SFN count at instant 1000 (Z) the service IP data flow is splitwith both unicast and broadcast RAT active ATSSS Release 16.




Other 5G use cases (not shown) using SFN count estab-lished in the core network and at UE can also be appliedto broadcast. This can be used to enable DisconnectedReception (DRX) on the UE, were the UE sleeps (quiescentstate) and awakes and becomes active at known SFN countinstant to receive, which saves battery on the UE. Also, as in5G IoT use cases, a device becomes active to receive a burstof data on a known SFN count cadence signaled by networkto increase IoT device battery life.

Moreover, by using the frame alignment and synchronizedSFN count in L1 signaling proposed, in both core networksand at the UE. The UE upper layers are then abstacted fromwhich channel unicast or broadcast used for IP flow fora service under ATSSS. This novel architecture for broadcast5G convegence proposed is a useful contribution for broad-cast and low-band spectrum in 5G for future research andconsideration.

Another synergy of NGWP is support for broadcast networkslicing, analogous to 5G network slicing [20]. The automa-tion of broadcast slice lifecycle is managed using theframework of Linux foundation Open Network AutomationPlatform (ONAP) project to align broadcast [15].

VIII. CONCLUSION AND FUTURE RESEARCH

This paper introduced a novel method for broadcast 5Gconverged services, aligned using methodology available inRelease 16. The proposed method shown was of a 5G MNOwith 5GC and BVNO using services of NGWP and with5GC and BC interworking using un-trusted non-3GPP ATSC3.0 RAT with dual connected UE using 3GPP Release 16.

Moreover, in USA a new 5G MNO has emerged in marketand has started to build out a new nation-wide greenfield 3GPPRelease 16 5G system architecture. The goal is to provide effi-cient utilization of low-band, non-contiguous broadcast spec-trum using innovative broadcast 5G heterogeneous networksand new business models.

Some of our future research is aimed at understanding thepotential use cases and value of time aligned frames withsynchronized L1 signaling and cooperating 5G and ATSC3.0 schedulers in context of ATSSS. Other areas of interestinclude use of artificial intelligence AI [19] and the LinuxFoundation’s Open Network Automation Platform (ONAP).The importance of security and interworking is recognized byauthors but was not discussed and is another central focus offuture research.

Finally, broadcast, Wi-Fi 6 and 5G convergence is beingstudied using similar methodology and a future industrycollaborative whitepaper on this subject is being planned.

REFERENCES

[1] “5G NR physical layer description,” 3GPP, Sophia Antipolis, France,Rep. TS 38.201, Dec. 2019.

[2] “System architecture for the 5G system,” 3GPP, Sophia Antipolis,France, Rep. TS 23.501, Mar. 2020.

[3] “Procedures 5G systems, stage 2,” 3GPP, Sophia Antipolis, France, Rep.TS 23.502, Mar. 2020.

[4] “Policy charging framework, stage 2,” 3GPP, Sophia Antipolis, France,Rep. TS 23.503, Mar. 2020.

[5] “Service requirements 5G system,” 3GPP, Sophia Antipolis, France, Rep.TS 22.261, Mar. 2020.

[6] System Discovery and Signaling, Standard ATSC A/321, Mar. 2016.[7] Physical Layer Protocol, Standard ATSC A/322, Jun. 2017.[8] Link-Layer Protocol, Standard ATSC A/330, Sep. 2016.[9] S.-I. Park et al., “Performance analysis of all modulation and code com-

binations in ATSC 3.0 physical layer protocol,” IEEE Trans. Broadcast.,vol. 65, no. 2, pp. 197–210, Jun. 2019.

[10] M. Fuentes et al., “Physical layer performance evaluation of LTE-advanced pro broadcast and ATSC 3.0 systems,” IEEE Trans. Broadcast.,vol. 65, no. 3, pp. 477–488, Sep. 2019.

[11] L. Zhang et al., “Layered-division-multiplexing: Theory and practice,”IEEE Trans. Broadcasting, vol. 62, no. 1, pp. 216–232, Mar. 2016.

[12] “Access traffic steering, switch and splitting support in the 5G system(5GS) architecture, Release 16,” 3GPP, Sophia Antipolis, France, Rep.TR 23.793, Dec. 2018.

[13] “Access traffic steering, switching and splitting; stage 3, Release 16,”3GPP, Sophia Antipolis, France, Rep. TS 24.193, Mar. 2020.

[14] S. Rommer, P. Hedman, M. Olsson, L. Frid, S. Sultana, and C. Mulligan,5G Core Networks: Powering Digitalization. London, U.K.: Academic,2019.

[15] J. Donovan and K. Prabhu, Building the Network of the Future: GettingSmarter, Faster and More Flexible With a Software Centric Approach,1st ed. London, U.K.: Chapman and Hall, 2017.

[16] Authorizing Permissive Use of the “Next Generation” BroadcastTelevision Standard (FCC). Accessed: Feb. 2018. [Online]. Available:https://www.govinfo.gov/content/pkg/FR-2018-02-02/pdf/2018-01473.pdf

[17] “Wireless and wireline convergence access support for the 5G System(5GS), V16.2.0 (2019-12),” 3GPP, Sophia Antipolis, France, Rep. TS23.316, Mar. 2020.

[18] “Solutions for NR to support non-terrestrial networks (NTN), V08.8.0(2019-9),” 3GPP Sophia Antipolis, France, Rep. TS 38.821, Dec. 2019.

[19] D. M. Gutierrez-Estevez et al., “Artificial intelligence for elastic man-agement and orchestration of 5G networks,” IEEE Wireless Commun.,vol. 26, no. 5, pp. 134–141, Oct. 2019.

[20] V. Q. Rodriguez, F. Guillemin, and A. Boubendir, “Automatingthe deployment of 5G network slices with ONAP,” 2019. [Online].Available: arXiv 1907.02278.

[21] L. Zhang, W. Li, Y. Wu, A. Prasad, S.-I. Park, and N. Hur, “Usingnon-orthogonal multiplexing for enhancing unicast-broadcast transmis-sion capacity in 5G,” in Proc. 2nd 5G World Forum (5GWF), Dresden,Germany, 2019, pp. 214–219.

[22] P. Klenner, J.-S. Baek, N. S. Loghin, D. Gómez-Barquero, and W.-S. Ko,“Physical layer time interleaving for the ATSC 3.0 system,” IEEE Trans.Broadcast., vol. 62, no. 1, pp. 253–262, Mar. 2016.

[23] 5G Wi-Fi RAN Convergence—Global Architecture and Policy,WBA, West Bromwich, U.K., Sep. 2019. [Online]. Available:https://wballiance.com/5g-wi-fi-ran-convergence-global-architecture-policy

[24] Managing 5G Converged Core With Access Traffic Steering,Switching, and Splitting: From Hybrid Access to ConvergedCore, IGI Global Company, Harrisburg, PA, USA, Jan. 2019,doi: 10.4018/978-1-5225-7570-2.ch008.

[25] E. Garro et al., “5G mixed mode: NR multicast-broadcastservices,” IEEE Trans. Broadcast., early access, Mar. 18, 2020,doi: 10.1109/TBC.2020.2977538.

[26] “Study on access traffic steering, switch and splitting support in the5G system architecture phase 2,” 3GPP, Sophia Antipolis, France, Rep.SP-200095, Mar. 2019.

Michael Simon received the M.S. degreein telecommunication from the University ofPittsburgh. He is currently the Director ofAdvanced Technology, Sinclair Broadcast Group.He has over 35 years of diverse experience inthe fields of broadcast engineering and wirelesstelecommunications and holds 20 patents. Hismain research interest is on development of a nextgeneration wireless platform based on SDN/NFVand using open networking tools. Which can enablebroadcast as a new 5G converged vertical by

interworking of 5G core and new shared cloud-native ATSC 3.0 broadcastcore and transmission infrastructure vRAN aligned with the 5G systemarchitecture starting in release 16.


http://dx.doi.org/10.4018/978-1-5225-7570-2.ch008

http://dx.doi.org/10.1109/TBC.2020.2977538



Ebenezer Kofi received the master’s degree ininformation systems from KNUST, and the Ph.D.degree in mathematics from North Carolina State.He is a versatile Solution Architect and DigitalTransformation Agent with commitment to ser-vice excellent. He lead Global Research andDevelopment Team related to IMS, 3G/4G/5G wire-less systems, leveraging machine learning and arti-ficial intelligence for 5G cloud native, multiac-cess edge environments (MEC), and end to endautonomous network with closed loop automation.

Currently focused on driving heterogenous network modernization effortsthrough innovative collaboration, delivering next generation open multido-main, event-driven, context-aware architecture (IaaS, PaaS, SaaS, NaaS,BaaS). He is a Member and Contributor to Opensource Community, 3GPP,Linux Foundation, and Open-Air-Interface Software Alliance.

Louis Libin (Member, IEEE) received the under-graduation degree in executive management pro-gram of optical electronics and high speed videog-raphy from MIT under Prof. Harold “Doc”Edgerton, the B.S.E.E. degree in physics fromYeshiva University, and the Electrical Engineeringdegree from Pratt Institute. He is the Vice Presidentof the Spectrum Policy and Strategy for One Mediaand Sinclair Broadcast Group. In 1996, he foundedBroad Comm., Inc., a consulting group special-izing in advanced television broadcast, interactive

TV, Internet protocol, and wireless communications. He has been anExecutive Director of the Advanced Television Broadcasting Alliance sinceJanuary 2014. He also served as the Senior Director for advanced technologyfor Sinclair Broadcast Group from 2015 to 2018. He was the Chief TechnologyOfficer for NBC and was responsible for all business and technical mattersfor satellite, wireless and communication issues for general electric and NBC.Since 1989, he has represented the United States on satellite and transmissionissues at the International Telecommunications Union, Geneva, Switzerland.He also serves on the boards of directors of several private and not-for profitcompanies. He completed a radio propagation software program at GWU,under Dr. Fred Matos. He was a Lecturer with the Walter Cronkite Schoolof Journalism on Media Entrepreneurship, Arizona State University for twoyears. He is a member of the National Society of Professional Engineers.

Mark Aitken (Member, IEEE) joined the SinclairBroadcast Group in 1999, where he serves as aSenior Vice President of advanced technology. Heis responsible for the formation of ONE Media,an innovation driven technology venture focusedon development of ‘Next Generation’ systems, anda driver of key ATSC 3.0 breakthrough technolo-gies, serving as its President since 2016. He hasbeen involved in the broadcast industry’s migrationto advanced services since 1987. He is a recipientof the 2008 “Broadcasting and Cable” Technology

Leadership Award, the 2013 ATSC “Bernard Lechner Outstanding ContributorAward,” the 2018 Awardee of “Future’s” Industry Innovator Award, and the2018 NAB Television Engineering Achievement Award. He is also a BoardMember of Saankhya Laboratory, an Indian based chip designer and leaderin SDR and SDN technologies. He is a member of the Association ofFederal Communications Consulting Engineers, the Society of Motion Pictureand Television Engineers, and the National Association of Broadcasters TVTechnology Committee which is focused on the broader technical issues ofour industry and an active participant in the ATSC.


atsc 3.0 broadcast 5g unicast heterogeneous network ......2020/07/13 · the 5gc architecture is in...

Documents